Using simulation to assess pre/post bed needs for an interventional platform concept
In this webinar, architect Frank Zilm discusses how simulation was used to explore the implementation of an innovative interventional platform concept at a new facility, as part of Saint Louis University Hospital’s $500 million renewal project.
Frank explains how the concept, which integrates surgery, cardiac procedures, interventional radiology and endoscopy services, can help minimize patient flow, improve efficiency and promote quality outcomes.
He discusses how the team approached the challenge of modeling and determining the optimum number of pre and post procedure beds for the interventional platform.
Watch the webinar to learn more about how simulation was an essential part of the project and how the technology facilitated discussion and decisions around changing the way key hospital services are delivered.
Frank Zilm is a registered architect with over 40 years of experience in health facilities programming and planning. He has published numerous articles and is a well-known speaker in the healthcare industry and applies simulation modeling techniques to evaluate complex operational and facility issues.
Frank has served as President of the AIA Academy on Architecture for Health and the Architecture for Health Foundation and was recognized with the Lifetime Achievement award in 2015 by the American College of Healthcare Architects.
Welcome to the session today. This is going to be a really interesting discussion presented by Frank Zilm. He is the Director of Healthcare Design, University of Kansas where he lectures. Relevant to this particular session, he also has 30 years of experience owning his own firm, Frank Zilm & Associates. As part of that experience, he has worked on over 300 planning studies [01.30]. He is quite experienced in this area.
He is also an active contributor to the healthcare community as a whole, including offering papers, regularly teaching at the Harvard Graduate School of Design’s executive programme. He serves as the President of the American Institute of Architects’ academy on architecture for health and he was also a founding fellow of the American College of Healthcare Architects. He has quite the resume. We are really excited to have him present to you today. [02.00]
With that, I want to turn it over to Frank. This is going to be a really great presentation and I think you really want to pay attention because it is a unique opportunity to explore how process changes and operations have been tied into the operational planning and building a facility that is going to support their needs in a way that was not historically able to be done in the existing building. It is also really innovative in that they are doing some really creative design work to build in flexibility [02.30] and the ability to grow the programme as needed.
Thank you, Brittany. Good morning and good afternoon depending on where you are located. I was going to try to get the webcam here working so I could at least wave to you and you can determine I am not a bot but my computer seemed to get traumatised by the idea of showing my image across the internet. I will not be able to do that. You will have to trust that I really do exist. [03.00]
As Brittany mentioned, I have had the opportunity to do a significant amount of simulation modelling, going all the way back to 40 years ago in batch mode with keypunch codes and IBM mainframe. Those of you that are familiar with simulation as a tool understand how robust and sophisticated it can be in dealing with complex problems, which, what I am going to show you today [03.30] is a way we built what I would consider some relatively simple models but use them in a way to help us deal with a very complex problem of scaling out the support areas for a new concept for a hospital. I am hoping that during our session today that Kelly Baumer from Saint Louis University will be able to join us. She is the Vice President for Special Projects and was a help person in our help [04.00] in making this a success.
II. Saint Louis University Hospital
Here is what I am going to try to accomplish [04.30]. We are trying to try give you a brief overview of the interventional platform and talk about the visioning process that we went through on this case study, the developing of the models, the output and then leave things open for questions and answers. I am assuming the audience is a mix of clinical professionals, system modellers and, hopefully, architects and planners. I am going to go over some basic fundamentals just to make sure we are all on the same playing [05.00] field in terms of understanding the same concepts and what we are going to try to do. Bear with me if I touch on something that seems fundamental to you. I am hopeful it will be of benefit to the other participants in this conference.
In the fall of 2015, the SSM system in Saint Louis reached an agreement with Saint Louis University to acquire Saint Louis University Hospital. The existing hospital [05.30] is shown on the left side screen. Typical of a hospital of its generation, it consisted of multiple phases. The initial tower was a 15 storey very iconic building called the Desloge building, which is still being used for patient care. There were a number of additions of varying age, quality and functionality. The commitment that SSM made to Saint Louis University was that they would [06.00] build and open a new hospital within a five-year time frame. Five years sounds like a long time but anyone who has been involved in the planning, design and construction process realises for the scale of hospital we are dealing with here – 300 beds and outpatient spaces – that is an aggressive timeline.
About a month after that agreement was made I was selected along with four other firms to lead [06.30] the initial planning of the project phase, working with a very strong leadership team composed of SSM and Saint Louis university physicians. The goal of that initial process was to basically get to a point where we would accurately scale all of the elements that would be involved.
The challenges were two-fold. One was that we were basically given a three-month timetable to [07.00] programme the entire complex, which was extremely aggressive and just happening to include Christmas and Thanksgiving holidays in it. It was thus three months minus some holidays where we could not really have client meetings. The second challenge was a very tight capital budget. We had to be looking at innovative ways to do things and also cost-effective ways to do things. [07.30]
Anyone that has been involved in a process – whatever the right term is – like this realises that if you look at behavioural economic studies and other studies of human behaviour, when someone is approached with a problem that is perceived as a high-risk scenario, we tend to become very conservative. It is human nature. The creation of a new complex opens us [08.00] the opportunity to look at new ways to do things. The challenge that we had was to bring to the table possible new operational and organisational models for Saint Louis University to look at and allow the users, team and leadership to have sufficient information and time to make good decisions. Not all new ideas are appropriate, but we wanted [08.30] to make sure that new ideas were explored and presented in a way that there could be careful evaluation and consideration of them.
The blue box, as you see in these diagrams, were workshops led by Kurt Salmon, who was one of the teams that were involved. The goal there was to look at the current process of the different units and develop possible alternative strategies with the renewable project [09.00] on how to approach these. We had roughly 12 team members working in this space with different components. The component I am going to talk about today is the interventional platform.
4. Interventional Platform
To make sure we are all on the same page, if you look at a traditional hospital you will see multiple components that are involved at some level of direct intervention with the patients [09.30]. The traditional component is the surgery suite which consists of operating rooms, support areas and, typically, what we would look at would be a patient prep in recovery areas for discharge patients, which are called phase 2: PA recovery, or post-anaesthesia recovery or inpatient hold. That would be one component.
Typically, under the same roof you would see a separate component that might have your interventional imaging components [10.00] typically run by the imaging radiology service. They would also have their own patient waiting, prep and recovery areas. You would have cardiology that might have cath labs, EP, and even a hybrid in this case procedure room, again with their own waiting, prep and recovery area and endoscopy, again, with procedure rooms, support space and patient care.
That is a fairly traditional model. It has some clear advantages in that each of the components [10.30] to a degree has complete control over the performance of their area. They also have to deal with complexities that can occur, for instance, if cases are running late in endoscopy and you are on your staff to run your recovery area until six o’clock, what do you do? How do you manage the patients and how do you deal with overall duplication of services in terms of even things like locker rooms and break areas? [11.00]
One model that has emerged over the past decade is called the interventional platform. The idea here is you are trying to build in adaptability and flexibility over time by making common room sizes wherever you can so that you can adjust the function of the room to the particular modality and design it in a way that you could logically extend it. Typically, when we are planning at this stage we anticipate what we call SDNT [11.30], which stands for Some Damn New Technology. Usually by the time the project is completed, there is something that needs to be accommodated that nobody had anticipated at the beginning of the project.
The other key element is the ability to consolidate prep recovery. I will talk a little bit later about what we are calling phase 2 areas in a way that you can ensure maximum effective utilisation of the space. For example, if cardiology [12.00] physicians go off Las Vegas for a conference and their workload drops down that those rooms would be available as prepping areas to other services to take advantage of. The idea is how to get maximum utilisation of the resources to position things in a way that you can deal with adaptation and change over time.
5. Case Study
A small-scale example of that is in a project I worked on a few years ago at University Medical Centre at Princeton [12.30]. I will try and see if I can draw here without causing too much trouble. If you look at this floor plan, this area is the operating rooms and this is the cath and EP areas. This area is the interventional platform. Supporting that, we have a common prep holding area and then a common PACU. I am sorry [13.00], I forgot my friends in endoscopy. They will probably offer me some free procedures for doing that, but endoscopy was also a piece of this.
This is a good case study because during construction it was determined that they needed to add a hybrid room in so they were able to modify this concept and adapt it to accept a hybrid room. All the other pieces worked. That is a small scale example of what an interventional platform is. [13.30]
What we had to do in terms of our work at this phase was developed what is called a space programme. That is to identify every square foot of space and the major functions that space would be involved in in the interventional platform at Saint Louis University. We are dealing with hundreds of rooms and areas that would need to be accommodated. This is the diagram I showed you earlier is the Desloge tower [14.00], a 15 storey iconic building on the campus. This was the building site that was being looked at. If you look right, you with see the massing concept for the replacement hospital. What we were going to deal with, in essence, was the first floor of that which would include the procedure rooms, interventional areas, staff areas and the different support areas associated with it.
The challenge was how to scale those support areas. [14.30] This is a diagram of what it was looking like in early schematics. This is the operating rooms, the interventional rooms, cath labs and endoscopy. These were the support functions for that, including PACU prep etc. I will be focusing on the discussion.
The challenge was how to scale this. There are different rules of thumb if you look at different areas that may or may not have validity. For example, the facility guidelines instituted in the United States [15.00] for operating rooms had a minimum standard of 1.5 PACU beds in their 2014 guidelines, which I have found in my research and past experience to be a very poor indicator of the true need for PACU beds, particularly if you look at the difference between a major tertiary hospital and ambulatory surgery facility. You are going to get radically different numbers. There were not good standards that we could identify [15.30] and, due to the time pressure that we were dealing with, with the three-month target, what we agreed to do with leadership was to put a placeholder in for our best estimate of the square footage needed for the support areas to meet that three-month target, so that the project could proceed in terms of validating the project budget and starting to develop the overall concepts. Then, shortly [16.00] at the beginning of this year, with a two-month target timeline, we initiated a process to validate the right mix of beds.
The process that we went through, I think, worked extremely well from my experience. What we did is we brought representatives from all of the components – cardiac, cath, imaging, radiology, imaging services, endoscopy and the surgery suite – together [16.30], we explained to them what we were trying to do in terms of trying to estimate the right number of post-anaesthesia recovery-area beds. We prepped phase II, and here were are talking about phase II being a lower level of observation of patients, and these would be in individual rooms; a PACU would typically be in an open bay area. Next is what we were calling a phase III area, which was basically a recliner area where patients with minimal anaesthetics [17.00] who were having an extended procedure that could be observed in an area would be managed through recliners. For the main hospital, this was a new concept this. The hospital had two separate surgery location sites, and the remote location site had some experience with this, but, for the main hospital, it was a new procedure for them.
The last piece was a short-stay [17.30] unit which would be a 23-hour observation unit, and I would argue, in my experience, that observation beds in the United States have been under-recognised as a component of healthcare delivery for a number of reasons, one of them being financial, but it is not uncommon for – and certainly, in this case or situation – observation patients – and I am defining observation patients as 23 hours or less – would be sent to a nursing-unit floor [18.00] to occupy an inpatient bed, which, in this case, was occupying a bed on a unit that are typically a very high census level. One thing that you see in many cases like this is, once the patient is moved to an inpatient unit, they start to be managed at a pace that is similar to other inpatient activity, as opposed to a short-stay unit, where, again, the goal would be to move that patient as quickly as possible.
One of the things [18.30] we were proposing here for this hospital as a new concept was, immediately adjacent to an interventional platform, we would have a short-stay area that would manage patient care of 23 hours or less. We wanted to be able to test the sensitivity of our results to what I call ‘what if we’re wrong scenarios’: how do we design in, to the degree we can, flexibility? As you will see in many cases, we are dealing with working assumptions and best judgment as opposed to hard [19.00] data about how the system will work in the future. We also wanted to be able to recommend back to leadership any adjustments to their preliminary space listing that we made. That was the task.
My favourite quote is this one and, again, I think the challenge is whether would we build models that would be effective in helping us make judgment decisions That is the challenge that we had.
III. Planning the Process
What we did is, again, after we met with each service [19.30], we then worked with the individual services to develop models of anticipated flow for the major patient types. We established a prototype schedules for patients. We then ran the models, reviewed those models with representatives from each service and tested the sensitivity. We were reviewed our findings with the staff from each service before presenting it to leadership, and then presented to leadership adjustments [20.00] that we felt were appropriate, based on our modelling and feedback from the users, so heavy user involvement – and very positive, in my experience, user involvement.
I will show you blow-up of this. It looks like an eye chart now but I will show you something that has a little more detail. For each service. What we did is, for each service, we did a Visio Flowchart of what they were doing today, and then we started to talk about, ‘If we have these components and a new facility [20.30], how would you anticipate using it?’ What we found is that, across the four major services, there were 19 different types of patient procedures that will be managed in there, so a fairly complex mix of different types of patients, each creating different patterns of demand on the pre- and post-procedure areas.
Just to take one – and I joke about this – close to my heart is electrophysiology and, as you can see here [21.00], what we are saying for our future model is that 60% of cases would be outpatient cases, 40% would be impatient cases, and there are slightly different flows for each of these patterns. For the outpatient cases, we are anticipating 60 minutes’ prep time, where, for the inpatient, we are assuming most of their prep is occurring in the nursing unit and there will be a shorter stay – primarily a holding stay – in the interventional-platform area. In terms of procedure times [21.30], any of you who are familiar with electrophysiology know that those procedures can go all over the place in length of stay, so we had a pretty broad distribution on the length of stay.
After the procedure for the outpatient area, however, 75% will be going to the short-stay unit, and that was a significant change in how they were currently managing patients, but they felt that that would be appropriate if that resource was immediately available to them. The others would go to phase II [22.00] and then they would be discharged. For the inpatient side, just to follow through on this particular example, 90% were anticipated to go back to a phase II area before going back to an inpatient unit. 10% were anticipated to acquire a post-anaesthesia recovery.
We did these models with every service and, if my memory is correct, we had at least [22.30] two reviews of the process. It was a very helpful step because, as they got more comfortable with what was being envisioned, many of the teams made some modifications, just saying, ‘We were evaluating this and, if we do have short-stay, we might change how we do things.’ That was step number one: we developed the process.
The next thing we did was to develop a prototype schedule and it was a fairly straightforward [23.00] [inaudible] use an Excel spreadsheet and we said, for example, what you are seeing on the left is the operating rooms, and there, rather than go into specific procedures like open-heart trauma, we divided the flows into short OR times, medium OR times and long OR times. As you can see, for example here, we are anticipating that six of the operating rooms would have a long case. It would be the only case done [23.30] in that room, starting at 6.30 in the morning, and that is a little unusual but they did have some fairly complex tertiary procedures, which has a significant impact, as I am sure you realise, on the demand for PACU beds.
Compare that to the short stay, where we are using five rooms and bringing them in in an hourly pattern into the system and making a break between morning and afternoon. Anybody who has tried [24.00] to model operating rooms will, I am sure, recognise that you really do not deal with a day schedule, but you deal with two half-day schedules. A lot of times, if you get a case cancellation or a break in the morning, it is difficult to move that afternoon case up in all cases, because it may be a different surgeon who has clinic in the morning and who does not come in until the afternoon. Modelling the scheduling of what happens in a surgery is a very complex problem. This is a fairly [24.30] simple way to try to resolve that.
One of the things we used was a scheduling module and SIMUL8 which allowed us to vary that time of arrival, trying to reflect the fact that you may plan to bring a patient in from the waiting room into the prep area at 9.15 but, if things are backed up, you may delay bringing that patient in for 15-20 minutes. Conversely, if a case breaks early, you might try to move the patient in from the waiting area into the prepare area a little bit ahead of schedule, or bring the [25.00] patient down a floor.
On the other side, you can see the cardiovascular, moving diagnostic patients through at a pretty fast 30-minute clip. Interventional, we are starting to stretch it out a bit, again heavily emphasising the morning electrophysiology [inaudible] TEE cases. We again did that for each of the 19 different flows that we had identified coming through the interventional platform.
IV. Developing the Simulation Model
[25.30] We did make some key assumptions in our modelling. We assumed that there would be no back-up in the surgery suite because of boarding issues and finding inpatient beds available. We are working on the assumption that our estimate of inpatient-bed needs would minimise any backlog into the interventional platform, so we are saying that the ability to move patients through is going to be primarily due to the resources on the interventional platform [26.00].
We also assumed no siloing of beds, in effect, and that all the prep and phase II beds would be shared, so that a service could not say, ‘These are our six prepare phase II rooms and, if we are not using them’, in essence, ‘too bad.’ They would be managed in a way that you get the maximum use out of the prepare phase II resources.
We did assume that inpatients would be handled in the [26.30] prepare area rather than the PACU area, which was a change of procedure for them. We looked at the mix of patients. As I mentioned earlier, one of the issues that you will see emerge was the demand for PACU beds, and we made different sets of assumptions. One assumption we made was that only 30% of short-duration surgery patients would be routed to PACU; the remaining 70% would go to phase II or to phase III. We also assumed that the turnaround times that we had would be built in – [27.00] the cleaning and the prep of the room.
This is a screenshot of the basic SIMUL8 model. Each of these icons that you are seeing in the screen is a sub-window. We have a process that I will show you for each component, but we identified them. Each area has its own sub-window, so you are not confronted with a mass of flow diagrams on the home sheet. Then we modelled the beds as resources. We have [27.30] different beds that we are saying are the resources that will be drawn on at different points in the process. One of the things I learned as I was working on the modelling is that at least the version of SIMUL8 I had has certain defaults built in to resources. They had travel time, for example, as a default value in the resources, so we had to go back in and adjust that to say, ‘That is not the way we are going to use resources in here – we are not going to have travel time.’ You would need to be careful to make sure that the [28.00] defaults on your resources are consistent with how you want to use the SIMUL8 tool.
Just to blow one of those sub-windows up, this is the interventional-cath area. At patient arrivals, we randomly distribute those patients based on the probabilities between outpatient, inpatient and same-day. You grab a prep area, do the procedure area and then, based on the flow charts we looked at [28.30] earlier, route which way the patients would go. What we did then is, as we approached an area – for example, the prep area – we would request the appropriate resource and, again, that would be, in this case, the prep II bed resource.
As I will show in a second, the strategy we were using on the prep and holding areas is that we wanted to model this as an unconstrained variable [29.00]. We did not want to be in a situation where the prep and the phase II beds were restricting the ability to use the procedure rooms, for a variety of reasons: 1) it is bad patient care; 2) it is underutilising the maximum resource you have; and 3) the last thing the CEO wants to have is irate surgeons storming into the office complaining about how bad the system is working, For those reasons, we said we want to be able to say, ‘We think we always have [29.30] enough beds to meet the maximum demand for those resources’, which means that, on average, we are probably not going to have full utilisation.
We also tried to insert, to the degree we could, variability in the procedure times, usually mostly average off of mean for some of the cases and, in some of the longer cases, such as the holding-bed scenario, we used a uniform distribution between six and 23 hours and said [30.00] the length of stay in the short-stay units fall somewhere in that. We introduced variability, both in the scheduling of arrivals of patients, the distribution of patients, and then the length of stays at different locations in the pre and post area.
This just shows a typical run that I am sure those of you who have done SIMUL8 before are familiar with. We are looking at time across the bottom in minutes. We are looking at number of beds used. This happens to be phase II beds. [30.30] If you could see the timescale a little more accurately, many of you would recognise what is going on here. First thing in the day, we are getting peak demand because we are filling up all the phase II beds in terms of prep for the different procedure areas. Everybody loves to start early. Typically, what you then see is a second wave that comes in around 11-12 o’clock, when we get the returning patients from the procedure areas coming back for phase II recovery [31.00] and we are moving the next wave of patients in for holding prior to procedure. We typically see a couple of spikes in the demand, and then the night time is across here.
V. Initial Simulation Results and Recommendations
We took the output from the demand and we translated it into Excel spreadsheets, where we looked at the cumulative demand for different types of beds. What we have on the vertical here is the cumulative demand. What you see [31.30] in different colour codes are the different types of beds. For example, we could say that, 90% of the time, we used 28 beds or less. Again, what we are looking at is the 100%. We are saying that we want to build this to the 100% level, because we do not want these beds to be a constraining variable. One could argue that you might be able to develop models that could move patients to different areas. We did not get into that simulation, in part because we said [32.00] that that is not the desired model, even though that might be what you would do if you were confronted with a capacity issue.
What we see across the top here was where we left things when we finished the initial programing. These were the number of beds we had placed in there. I see that Kelly Baumer has joined us as a panellist here, and she can provide more insight on this. I have to confess that, in my opinion, I [32.30] probably oversized these areas. I know I oversized them, based on my judgment. Again, we did not have hard data to verify that, but what I did not want to have happen is, when we came back with our refined numbers and the numbers we felt accurate, that we blew up the budget. My hope was that what we would come back and report would be something smaller than what we had originally put in there as a placeholder at the end of the programing stage.
[33.00] What you see across the bottom was our first initial results, where we said we needed 40 phase II areas, 11 PACU, 14 phase III and 16 short-stay. A couple surprises came out of this. One was that the PACU number made the PACU nurse extremely nervous. I think that might be an understatement. It was a surprise to all of us, so we went back and looked at that. This number of short-stay beds was interesting because I really do think it was reflecting [33.30] some forward thinking by everybody involved about how care will be provided in the future. This was a bigger number than we had anticipated, but I think it really reflects a shift in care that is an appropriate assumption about how to work in the future.
I just have a quick question for you. You said earlier that there are standards about ratios: 1.5 beds per [34.00] [inaudible]. How do these numbers compare to that? Do you know that off the top of your head?
No comparison at all. For example, there are 14 operating rooms, so, if you use the FTI guidelines, you would be talking about 21 beds. FTI guidelines and [inaudible] guidelines are very good overall, but I think this is one area where, in my judgment, they dropped the ball because [34.30] setting a standard like that does not reflect the diversity of types of operating rooms we are talking about here. We are talking about a very low percentage of low-acuity patients and a very high percentage of long-procedure-time patients, which then allows you to clear out the PACU. I would say, in most cases, our numbers were coming out under the rules of thumb that you might encounter as you look through the literature.
[35.00] We then did some sensitivity analysis of our initial assumptions. We said, ‘What if 70% of the short-stay patients went to PACU rather than 30%? What change does that make?’ That bumped that number up a little and bumped this number down a little, because we were not taking them to the phase II area. We said, ‘What if 70% of surgery short-stay and all long cases went to the PACU?’ and then that bumped that number up. [35.30] This number stayed the same, which was a little surprising and, again, you can see the short-stay numbers were holding constant, because we really did not vary that in the initial sensitivity analysis.
We reviewed all of this with representatives of each of the services and took them through how we model this. Again, we just looked at different sensitivity. Basically, we got sign-off on the following adjustments to the original [36.00] programme: we were going to shift six beds from the phase II area over to short-stay to meet that 16-bed target, and then we were going to shell beds to allow us to deal with potentially future growth or ‘What if one of our assumptions is wrong?’ We are basically saying we will end up with 44 prep/phase II beds and 16 short-stay beds but, of the 44, you would have some shell. We would shell PACU beds. Originally, we had 16 beds in our [36.30] programme; we said we will go with 12 initially and shell, and then we reduced the phase III area, which is the recliner area, to 14 stations.
That was presented to leadership within that two-month timetable and accepted with, I would say, minimal questions. I think people understood. The key thing is people understood how we attacked the problem [37.00], how we approached it, how we analysed it, what the models were saying, and then what adjustments we were making.
Kelly Baumer, who was an absolute key person in this process, is Vice President of Special Projects, and she has been working at St Louis University since 1997. She had extensive background knowledge of how things worked and a great deal of trust by the team members. Kelly, I do not know [37.30] if you want to make any comments about your observation of this process or anything that might have transpired since we completed the modelling work.
The thing I think I just want to add to all of the things that Frank has said today is that this modelling process was really crucial to getting us to [inaudible]. As you look at the models that Frank was presenting, you can see that we have a great deal of variability in procedure times and in the complexity [38.00] of work that we do within our cath labs and in all the various interventional areas. Before we did this modelling, it made people really nervous to say, ‘Are we building this PACUs and pre- and post-op areas correctly?’ It took a lot of work by the team and it really does take engagement, not only by your leaders but really by your frontline staff who are doing the work every day. Frank provided great spreadsheets for us [38.30]. We took a lot of time to go through and model it. Once we showed the work groups the product of the models and how we came to it, our physicians and our staff feel more comfortable. We feel more comfortable being in charge of the project, and it was then able to get the team that needed to sign off on the space comfortable that we did what we needed to do [inaudible] everything. [39.00] I just want to say that I think it was a great process for us to go through and it really made a huge difference in our comfort level about how we were sizing the space and where we were putting the beds etc.
If you would feel comfortable talking about it, my impression is that, from the physician perspective, there was some continued concern about the interventional-platform concept. My understanding is that you [39.30] have done a number of site visits to try to better familiarise the team with how interventional platforms have been operating at academic medical centres.
That is correct. We visited three different academic centres and they have done reconstructions. We have another visit [inaudible] who have done the same concept. As Frank told you at the beginning, our situation currently is that we have an outpatient [40.00] OR and four outpatient ORs about a quarter of a mile from our campus, where a lot of our outpatient work is done currently. The biggest concern that we had was bringing those inpatient and outpatient ORs into one platform. That was probably the biggest concern that our surgeons had. A little bit also [inaudible] interventional platform to all of those areas – endoscopy, AR and cath – together.
The biggest concern [40.30] really was from the surgeons’ side because we are an environment where we do level I trauma and transplants. The concern is that, if you bring these outpatients with your inpatients, they are constantly get bumped for these emergency procedures that have to go into an OR suite right away. We spent a lot of time working with facilities that have done this successfully on the process and how they have allocated those rooms. We learned [41.00] some things about fast-track processing and things like that, but our physicians are now more comfortable moving forward with this integrated platform. That is something that I would encourage. We went to facilities that were just like ours, that had the transplant and the trauma, and the surgeons there were saying that this model works. It has helped to get us [inaudible] moving on with our design.
Thank you both, Frank and Kelly, for your insights. Kelly, can I ask a follow-up question? You said you really thought this process worked well. Is this a process around timing that you had done before or was this a new approach that Frank was bringing to you? Do you have any reflections on that?
No, this is a new process for us. Again, as we went through our planning phases [42.00], we talked about sizes and beds and what our procedures looked like. What we quickly realised was, ‘What about the EP patient who takes 300 minutes versus those who take 100 minutes? What about the neuro-interventional patient who comes in emergently?’ Whatever procedure we were doing, there were so many variabilities in the complexity. We are a tertiary and quaternary centre, so we [42.30] do a lot of complex, long procedures, as Frank alluded to. Sometimes we may have a neurosurgery case that is in our OR for 12 hours and takes up one room, but then we have other rooms that turn over every two or three hours, so that great degree of variability left us with a lot of discomfort around whether we were sizing the PACU right and how we were doing our pre- and post-op beds. Now we are bringing the outpatient platform into the inpatient platform, how do our ophthalmology patients recover? All of these things presented so many variables [43.00], and we said, ‘We do not know if we are sizing this correctly.’
That is when Frank really recommended that we did a deeper dive into each of these specific things. Once we saw the models working through, it really made us all the difference in us feeling comfortable with the decisions we were making moving forward. We were on a very tight timeline with this project. Timing wise, it was helpful because we could continue moving [43.30] into the schematic-design phase, because we did not have a lot of time to wait two months to understand this. Putting these models together and letting everyone see how they flowed, we got to a point where we were comfortable with how we were moving forward.
That is great to hear. We had a couple of other questions that have come in from the audience, so I want to see if we can [inaudible] if we can. One of the questions [44.00] was to you, Frank: ‘Do you think that every project like this needs to be simulated individually or are there some key learnings and some ways that we could look at those guidelines and re-evaluate them if you are doing interventional platforms at an academic centre versus an outpatient centre? Could we establish some standards and improve those as opposed to doing it individually for each project? How do you think that works?’
That is a good question and I would say, if [44.30] there was a large enough case-study database, we could be doing some things. For example, if you look at the emergency-room planning, there is an alliance of emergency rooms that annually report statistics and data that you could start to get some basic descriptive statistics that might alleviate some of the questions that we were wrestling with in the interventional platform. My experience and knowledge is there are not enough of these in [45.00] play, and they are relatively different in scope. I do not know to what degree you could develop and effect a generic model or a generic set of recommendations that would be appropriate for all the different settings that we are dealing with.
When you insert the element of an academic medical centre, one academic medical centre is just like one academic medical centre. They are all [45.30] zebras and different, and so I would be, at least at this stage, pessimistic. Now, maybe down the road in the future, if we have hundreds of these that we could start to aggregate operational and demand statistics, you could look at trying to establish some basic guidelines but, in my judgment, I do not think we are there yet.
It sounds like your future research project. [46.00] Another question from the audience is looking at how the long procedure times allowed you to clear the PACU and reduce the number of beds. The question is: ‘How did you do some sensitivity analysis around how many people are going to be in those short procedures versus long procedures? What if new technologies come in and maybe shorten procedure times? [46.30] How does that affect the need and what kind of flexibility do you have along that?’
Again, that is a great question. We worked with the surgery-suite leadership to develop the prototype schedules. Now, I will say that, while we varied arrival times and we had variation in length of stay and in terms of [47.00] the routing – particularly the short stay – we did not do a lot of sensitivity analysis on the assumption that six of the rooms would be long-case rooms. That might have been something that we could have gone back to and perhaps, if time had allowed, tested.
Just thinking off the top of my head, which is always dangerous for me, if we are talking about the difference between an eight-hour and a six-hour procedure, I would argue that, typically, that is not [47.30] going to have a significant demand on the PACU area because, usually, most PACUs peak out late morning and early afternoon. As the workload starts to taper off, their demand drops down. If we had a case that went six hours rather than eight, or if the average changed, I do not know if that would have a dramatic impact.
If you remember, in our recommendations we did put in some shell space [48.00] just to say, ‘What if we are wrong? What if things go in a different direction?’ I think everybody on the leadership team recognised that, although they were on an extremely tight budget, that was a reasonable strategy to try to keep the interventional concept but buy some flexibility if the original modelling assumptions did not turn out to be right.
The last question here [48.30] from the audience is: ‘Rather than looking at the number of required units’ – I think maybe they mean beds in that case – ‘how do you consider using simulation to design a physical layout of the units using simulation as opposed to just looking at your total capacity? Have you done that before? What is that like?’
Our particular role in this case was to develop [49.00] the space programme, so we did not have a floorplan to work from. That would be a logical next step, particularly in emergency-room designs, for example, where you have hundreds of patients flowing through a particular layout. Looking at that flow and evaluating how that is working in terms of basic traffic and implications on access to different areas, I would argue, is an important consideration. That [49.30] would have been a logical next step as the project moved into schematic design. The architects involved did not choose to do that. They did some other techniques to look at flow and layout, but that would be a logical next step in this kind of a modelling exercise.
I have a question for you as well. Kelly, this is also for you. [50.00] Did you come forward to this project with, ‘Yes, we are doing interventional platform and it is going to be dynamic and we are going to build in some of this flexibility’? Did you come forward already having that [inaudible] in place, given your short timelines, or was that on the table in the sense of, ‘Does it make sense to put all the pre and post beds together?’ Was that a question you were evaluating or was it pretty well established when you kicked off the project?
Kelly, let me take a first shot [50.30] at that and then you can correct me if I say something profoundly wrong. At the beginning, we established taskforces for the different areas. We established a taskforce for interventional services, but we did not come in saying, ‘We are going to do this.’ We proposed this As one option and worked with them. Again, I would argue that one of our roles [51.00], as being the outsiders, is to bring in ideas from other areas and from other experiences and to say, ‘Do we think this would work in your case?’ That taskforce group, which, again, had representatives from all of the different services involved, basically concurred that the interventional-platform idea was appropriate.
Now I am not sure, and Kelly may be able to outline this, but my suspicion is that not all [51.30] of the physicians are 100% convinced that this is the right way to go. That is not unusual in this kind of a planning process. If you can get 100% buy-in to a concept, it is a very rare case and you ought to go out and party for a couple days, if you can do that. I think there is still some concern. I would not be surprised if a lot of it has to do with about the ability to control your own processes [52.00] and make sure they work right. Kelly, I do not know what your observation is.
There is a lot of concern on the non-OR side in terms of cath lab, AR and endoscopy. Right now, these are done in various floors in our hospital. We only have prep and recovery areas. Some of these procedures that are very undersized, we were constantly [52.30] having patients in hallways, so I think that they feel that a combined platform really makes a lot of sense. I think that the biggest concern that I still see from doctors is the combination of the inpatient and outpatient ORs in one platform. That really is the one concern that still remains. We have done a lot of work and have visited facilities to see how they have done it successfully, and we are really starting to get through that. [53.00] Outpatient surgeons are most concerned with being bumped because of inpatient emergent procedures, which is part of our whole business here. I would say that that is probably the one remaining thing that we are still getting surgeons to a comfort level with, but I feel like the rest of the interventional physicians are really pretty happy with how the process is moving forward.
That is great to hear. Is this something that you learned from your [53.30] site visits? Are you able to share with us where the visits were?
We visited Stanford. That was a little more about the outpatient side. We visited the University of Utah, which was on the ambulatory side in the inpatient hospital. Stanford is in the process of building a new hospital but they just built a new outpatient facility for ambulatory care. The University of Utah [54.00] is in the process of adding to their hospital, so we learned a lot from them. They are still in that process.
The platform that we looked at that was very helpful was the Zuckerberg San Francisco General Hospital, whose physicians are academics from the University of California San Francisco. They developed a very successful fast-track process especially for outpatient orthopaedics, whereby the concept [54.30] is that there are two OR rooms that can never be bumped and are only for outpatient procedures that are 90 minutes or less. They have been able to double the amount of surgeries that they do for outpatients in a day. They have a dedicated team of people.
The way in which it works is process-driven as well as design-driven. They have an extra circulating team. As one team is in the room doing the procedure, the other team is prepping [55.00] and doing everything they need to do to get the next patient ready. As soon as patient one is finished, that team brings that patient out to recovery and the second team is ready to go into the next room. They have turned their turnover time to almost negative numbers and it has been very successful for them. We learned a lot about that, and that is something that we are working on in our current facility and hoping to take to the new facility.
That is fantastic – really interesting. [55.30] I do want to wrap things up here, so thank you, Frank, for the presentation. Do either of you have anything else you want to mention before we close out?
I would just like to thank SSM for allowing us to share this material with you today. As I mentioned at the beginning, I think one of the keys to the success was having a strong leader from the project to work with the group, and I appreciate Kelly [56.00] serving in that role.
Thank you. The one thing I just want to add is that Frank and his team were really good listeners. I would say that really engaging your frontline staff in the process is really key, because they are the ones who know all the current challenges that they face and they always bring a lot of things to life that sometimes those at the leadership level are not thinking about. Additionally, you need architects who really sit and listen [56.30] to what the staff say, so that you can have successful models. I would say that those are the two key things that we were very lucky with in this project.
Thank you both so much. I really appreciate you taking your time. It has really been fantastic to hear about this project. What is really important is that we remember that simulation is part of an integrated project. It is not the focus of the project; the modelling is not the final conclusion. It is the technology that facilitates the discussions [57.00] and the decision-making and that helps you make those plans better. I really appreciate you putting it in that context for us today. Thank you, everybody, so much for your time and we will talk to you soon.